5
$\begingroup$

I have a histogram of points with a dip in the center of the bell, seeming to create two bells, or two clusters. What is the name for this kind of shaped distribution? This curve should be normal but doesn't quite seem to be.

histogram

$\endgroup$
  • 3
    $\begingroup$ What's the data? Perhaps there's a confounding variable that could explain the bi-modality? $\endgroup$ – naught101 May 13 '16 at 3:06
  • $\begingroup$ @naught101That does seem to be the case. I am investigating. The data is Ted Cruz pct vote win by county in the US primaries. $\endgroup$ – Learning stats by example May 13 '16 at 3:07
  • $\begingroup$ In that case it's probably worth mapping the data.. $\endgroup$ – naught101 May 13 '16 at 3:12
  • 3
    $\begingroup$ If the data are percentages of votes, then I don't see why they "should be normal". Different sociodemographics probably come with very different voting patterns. The overall result could have pretty much any shape. $\endgroup$ – Stephan Kolassa May 13 '16 at 8:04
  • 1
    $\begingroup$ "political data" might have almost any distribution depending on what variable you're measuring if you mean things like proportions of people with some political preference ("do you approve of candidate X?", like the ) perhaps across some set of locations (like your counties), I answered that already -- it's bounded and discrete; more specifically a finite mixture of discrete distributions (with rational mixing proportions). That's about all you can say. It might in some cases make sense to approximate it as a mixture of binomial distributions. That would be the first thing I'd try. ... ctd $\endgroup$ – Glen_b -Reinstate Monica Jun 26 '17 at 23:21
11
$\begingroup$

It could be a bimodal distribution

Or then it could just be the a run of the mill normal distribution as the dip in the middle doesn't appear to be that big.

bimodal distribution

Image Bimodal.png by Maksim, from Wikimedia commons; licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ How do we translate S.D. values into percentage and something like that results in bimodalty following a mapping of data. $\endgroup$ – Subhash C. Davar May 13 '16 at 5:01
  • 4
    $\begingroup$ @subash "Bimodal" simply means "has two modes" -- there's more than one bimodal distribution, so if I understand what you're saying there (it's somewhat unclear) then your question doesn't have a unique answer. $\endgroup$ – Glen_b -Reinstate Monica May 13 '16 at 7:02
  • $\begingroup$ The image from your post appears to be taken from wikimedia commons, authored by user Maksim (it was used in the Wikipedia article "multimodal distribution"). Both the conditions of the license and StackExchange's own standards require proper attribution when you use work like this. Please attribute the authorship or replace the plot. (You're free to use this one I just generated - i.stack.imgur.com/RdFuk.png ) $\endgroup$ – Glen_b -Reinstate Monica Feb 3 '17 at 2:28
  • $\begingroup$ @Glen_b - thanks for the heads up. Does the current update confirm with the standards? I'm not sure if I should attribute the article or the file itself. $\endgroup$ – plumbus_bouquet Feb 3 '17 at 3:07
  • $\begingroup$ @plumbus Thanks. Maksim made the picture, he didn't write the article, so it's it's his work making the image that you recognize ((though the addition of a link to that wikipedia page is useful) The first link I gave is the thing you're using (the wikimedia commons image) and that page contains a link to the license (actually the image's page seems to list two different licenses, you're presumably able to pick one). You didn't modify the image so the license requires only two things: give credit and link to the license. ...ctd $\endgroup$ – Glen_b -Reinstate Monica Feb 3 '17 at 23:54
1
$\begingroup$

A bimodal distribution. You could also say it's an almost-normal curve with negative kurtosis. (Kurtosis refers to the spikiness of a normal curve; a bell-shaped curve that is very tall and elongated in height would have positive kurtosis). Your curve also appears to have a long right tail, so it is skewed to the right.

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.